Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
APPLICATION OF FRACTURE
MECHANICS TO CONCRETE
Summary of a series of lectures 1988
Arne Hillerborg
GODEN: LUTYDG/(TYBM-3030)/1-28/(1988)
APPLICATION OF FRACTURE
MECHANICS TO CONCRETE
Summary of a series of lectures 1988
Arne Hillerborg
ISSN
0348-7911
REPORT TYBM-3030
LUND SWEDEN 1988
with
Introduction
Conventional fracture mechanics is mainly based on the theory of
elasticity and it is used for studying the stability and propagation
of existing cracks. The modern form of application of fracture
mechanics to concrete, which was first developed in Lund, differs
from the conventional fracture mechanics in both these respects.
In the basic form of conventional fracture mechanics it is assumed
that the stresses and strains tend towards infinity at a crack tip,
Fig. 1. This is of cours e not realistic. In spite of this the
theoretical results based on this assumption in many cases lead to
realistic conclusions. In other cases this is not the case. This has
since long
b~en
without any initial crack, whereas the modern mode l can also be
applied to this case. It is now possible not only to study the
stability and propagation of a crack, but also its formation. Thus
the complete development of fracture can be analysed by means of one
model as a continuous process. This is an expansion of fracture
mechanics, which opens quite new possibilities to analyse fracture
of real structures, as will be demonstrated by means of examples.
This circumstance has been emphasized here, because it has not
always been fully appreciated. Thus for example the new approach is
sometimes described as a variant of the so called Dugdale-Barenblatt
model. Dugdale and Barenblatt never realised the possibility to use
this type of model for the analyses of uncracked structures. They
only intended their models as explanations of the stress situation
in the vicinity of the tip of an existing crack.
Conventional fracture mechanics
As a background for the understanding of the new model a short
description will be given of conventional fracture mechanics, particularly linear elatic fracture mechanics, often written LEFM for
short.
If a stress is applied perpendicular to a crack with a sharp tip,
the linear elastic solution shows that a stress concentration will
appear at the tip, such that the stress approaches infinity, Fig. 1.
Close to the crack tip the stress distribution is approximately
described by the equation
/27rx
In this equation x is the distance from the crack tip and K is called the stress intensity factor. This factor can be calculated from
the equation
K = Ya/a
O" y-
V2Ttx
4
between a stress and a strength as fracture criterion. This means
that different models have to be used for cracked and uncracked
material, with different fracture criteria and two different
material properties, Kc and strength. This lack of continuity is a
drawback, at least for materials like concrete.
An alternative treatment according to LEFM is based on the stress
release rate when a crack propagates, i e the amount of energy which
is released in the structure per unit crack area when the crack propagates. An energy release rate G is theoretically calculated. The
crack i assumed to propagate if G reaches a critical value, the critical energy release rate Gc ' which is equal to the amount of energy
that is absorbed in the fracture zone per unit area when the crack
grows. It can be demonstrated that the approaches by means of stress
intensity factors and energy release rates are equivalent, and that
they are coupled by means of the following relation for the case of
plane stress conditions.
It has long been recognised that the unrealistic assumption that the
stress and strain approaches infinity can lead to erroneous results.
Many methods are used to make corrections in order to take into account the limited strength of the material. These methods will not
be discussed here, as theyas a rule give results, which are not accurate enough for concrete structures of normal sizes.
=
._
~.-r..:======'l==;;='.'i, =:J=====~2~
c
test.
The test is assumed to be performed in deformation control, which
means that the deformation is slowly increased. During the first
phase of the test, the stress increases as the deformation increases. This is said to be the ascending branch in the stress-deformation diagram. If the specimen is assumed to be homogenuous, the
relative elongation during this phase is the same along the whole
specimen. This means that the deformation can be described by means
of a strain
E,
length. The same stress-strain diagram is valid for the whole specimen at this stage.
Af ter the peak stress has been reached, the post-peak stage, a further increase in deformation means that the stress decreases. We are
now on the descending branch in the diagram. The cause for the decreasing stress is that the damage (microcracks) somewhere along the
bar has become so high, that any increase in deformation leadsto a
decreasing ability to transfer stresses. Within this damage zone
fracture zone an increase in deformation takes place, at the same
or
El + w
~-E-curve
~-w-curve
Fig. 4.
Material properties.
The stress-strain curve in tension for ordinary concrete deviates
rather little from a straight line. Thus for most practical applications this relation can be assumed to be a straight line. All applications so far seem to have been based on this assumption.
The stress-deformation curve corresponding to the descending branch
can be measured by means of modern test equipment. The shape of this
curve is now relatively weIl known. It has the general shape shown
in Fig. 5. One important parameter of the curve is the enclosed
10
~-w-curve
for
concrete.
50
100
150 w[pml
area, which equals the fracture energy GF Another important property may be the initial slope of the curve. This property and its
practical significance has not hitherto been studied in detail.
The fracture energy GF can suitably be determined by me ans of a
simple bending test on a notched beam according to a RILEM recommendation, Fig. 6. The load-deformation curve is recorded in the test.
The total energy, that is absorbed during the test, is equal to the
sum of the areas Al' A2 and A3 . Al is measured in the diagram, A2 is
calculated as F 1 S 0 , where Fl is a central force giving the same
bending moment as the weight of the beam and the loading equipment.
A3 is assumed to be equal to A2 . This total energy is divided by the
area b(h-a), that has been fractured, in order to get the value of
GF
11
For normal concrete qualities the material parameters are within the
following ranges:
E
GF
ft
Ich
=
=
=
=
20
- 40 GPa
65 -200 N/m
2 - 4 MPa
0.1 - 1 m
The value of Ich can be expected to be lower for high strength concrete and for light weight concrete than for ordinary concrete, which
means that these materials are more brittle.
For fibre reinforced concrete the shapes of the IT-E- and IT-w-curves
may differ much from those for plain concrete. These curves then
have to be determined and introduced into the analyses for the
particular materials in question.
Application, principles.
Wherever tensile strains appear, which tend to pass the strain
corresponding to the peak point in the tensile stress-deformation
curve, a fracture zone starts to develop. Then the model outlined
above can be applied, which means that the IT-w-curve is applied to
the additional deformation in the fracture zone.
Let us as an example look at the bent beam in Fig. 7. At low loads
the simple beam theory can be used, which me ans that the strains and
stresses vary linearlyacross the section, with a 'maximum stress
equal to
where M is the acting moment, b the width and d the depth of the
beame When the maximum stress reaches the tensile strength f t a
fracture zone starts developing if the deflection of the beam is
increased. As the deflection increases, the fracture zone grows into
12
2
6M
bd2
ff-t--~t--..
f t-
3
Fracture zone depth. af
13
14
15
where
El
E and
w are in
16
has been given, the smeared approach will not be discussed any further in this paper.
6Mjb(d-a)2
17
The value of the ratio ff/f t can be taken from diagrams of the type
shown in Fig. 7, and the ratio fnet/f t
from Fig. 8.
plastic
!)M
1----l1
ff~Mu/(bd2/6)
elastic brittle
tensile strength.
0.01
10
0.1
d/Ich
trift
2.5
Initial stress at
mid-span:
2.0
1.5
1.0
0.5
O
0.01
0.1
10
18
f net
ft
'3
2.5
2
100-w:(CJI
1.01.----
\\ lEFM
Q.:0.6\ \
1.S
\\0.2
4.0
\ \
\ \
a.o
\ \
\ \
\\
I(J-w. (el I
0.1 L-rrrrr--r--r-r-.,.....,.T'T"-~~Y---2
O.OS 0.1 02
0.5
o OL-L-L-L-L-~'-'-'-.l-.l-_
bending strength.
0.5
1.0
old
19
In all the figures it can be seen that the strength depends on the
size d of the beame Thus there is a size effect, which is explained
by means of the fracture mechanics approach. The size effect is
greater for notched beams than for unnotched, particularly for deep
notched beams. For unnotched beams this size effect increases when
shrinkage or temperature stresses are acting.
In Figs. 11 and 13 it can be seen that the strength approaches the
value predicted by the theory of plasticity for small beams, and
that it approaches the value predicted by the theory of elasticity
for large beams. The analysis covers all cases between these
extremes for unnotched as weIl as notched beams. It thus has a
general applicability. A small value of the "brittleness number"
d/Ich gives a more plastic-tough behaviour, whereas a high value
gives a more elastic-brittle behaviour.
Sensitivity analysis.
The formal strength according to the above results depends on the
value of d/Ich' where Ich in its turn depends on E, GF snd f t according to the relation Ich = EG F/f t 2 The diagrams are given in logaritmic scales (except Fig. 12). For a small change in d/Ich the
relation can approximately be written
df f
dft
dd
dE
dG F
(1-2B) f t - Ber + BIf + B GF
20
-------~-------- ~cr
l'
----~----
~)
21
ffi f t
Crushing failure
3.5
3
2.5
2
Fig. 16. Yariation
of formal bending
strength of a pipe.
1.5
0.1
0.2
0.5
10
dj/lch
22
0.4
~~
Fig. 17. Theoretical ratio
0.3
o
1.5%
1%
0.5%
0.2
reinforcement.
~.1%
L/d-6
0.15
S-l% Lld-9
0.1
0.2
la
5.0 d Ilch
23
The results of the analyses are shown in Fig. 17. The variables have
been the depth (expressed as d/Ich)' the span to depth ratio, and
the reinforcement ratio
9.
the beam depth, which is wellknown from tests, but which has not
earlier had any rationaI explanation.
A very large number of tests have been performed regarding the shear
strength of beams, and as a matter of fact all our knowledge, as expressed in building codes and text books, is based on these tests.
As we now for the first time have a pure theoretical analysis of the
shear failure, it is interesting to compare this with test results
and with the code formulas. Such comparisons are shown in Figs. 18
20. As all the material parameters in the tests are not known, particularly not the fracture enegy GF , it has on ly been possible to
make relative comparisons, which means that all curves in a diagram
have been drawn through one common point.
From all the figures it is evident that the theoretical results are
in a good agreement with the test results. Regarding the influence
of beam depth and reinforcement ratio they are also in a good agreement with the CEB Model Code, which is mainly based on the same test
results. The ACI code does not show any good agreement with the
theoretical results or the test results.
In Japan (Iguro et al, 1984) a test series has been performed, where
the beam depth has been varied between 0.1 m and 3 m, i e by a
factor of 30. From these tests it was concluded that the shear
strength is inversely proportional to the fourth root of the depth.
This corresponds to a 45 0 slope in Fig. 17, and it is thus in a good
agreement with the theoretical results.
Based on the theoretical analysis and the test results it can approximately be assumed that the following equation is valid for a
beam with a constant span to depth ratio and a constant re in forcement ratio:
24
~ /Vu. d =O.2m
1.4
1.2
ACI
1.0
0.8
CES
............
,-
0.6
--_____
......0
"c
1
0.4
0.2
El
0.2
0.4
0.6
0.8 1.0
1.2
Seam depth, d(m)
fe ..3000psi
o
O
9 lid
to depth ratio.
beam depth.
~-CEB
1.6
1.4
ACI
1.2
1.0
f~=3000 psi
0.8
---
oO
Colcul'otions lid
0.5
lO
=3,
1.5
2.0
ced beams.
2.5
Percentage reinforcement,
Fig. 20. Influence of
reinforcement ratio
9(%)
25
From this expression it can be seen that the shear strength depends
as much on GF as on'ft 2 It is generally accepted that the tensile
strength f t
the compressive strength. The conclusion from this is that the shear
strength of a beam depends as much on the fracture energy of the
concrete in the beam as on its compressive strength.
When a laboratory test is performed on a concrete structure, the
compressive strength is traditionally always measured and reported.
From the above it follows that the fracture energy should also be
measured and reported where shear tests are performed, as this property is as important as the compressive strength. The same may hold
also for many other types of structural tests.
AIso in code formulas for shear strength the fracture energy ought
to be taken into account in some way or other. How this should be
done is too earl y to specify, but one possibility could be to give
some type of correction factor, depending on the type of concrete,
for example reduction factors for light weight concrete and for high
strength concrete.
26
27
a reinforced beame These first results indicate that the stressstrain relation to be used for the practical design should preferrably have an ultimate strain equal to kl/x, instead of the normally
assumed 3.5 permille, where kl is a material property and x is the
depth of the compression zone. If this conclusion is correct, it
will have a significant influence in many practical situations.
Further research is needed before the result is sufficiently confirmed.
In a third project a number of tests are being performed on some
simple unreinforced structures in order to check the general applicability of the fracture mechanics aproach. A wide range of different materials are tested, particularly with respect to different
toughness. Very brittle materials are tested, like pure cement paste, as weIl as wery tough materials, like fibre reinforced concrete,
and some materials in the intermediate range.
3.00
~1.00
Z
o
2.00
vi
ffi
~
;.BOO
1.00
..J
<
~
z~
..J
~.600
.000
Z
.~oo
-1.00
.200
-2.00
.000
.200
.~OO
.600
.BOO
i.OO
i.2O
i.~O
-3.00
SHEAR DEFORMATION. mm
.000
.200
.~OO
.600
.BOO
1.00
NORMAL DEFORMATlON.mm
~ 6.00
~MO
er
~
~ 3.00
2.00
1.00
I
I
I
.000
.200
.0400
.600
.BOO
1..00
1.20
1..040
Sf-EAR DEFORMATION. mm
28
References
American Concrete Institute (1983) Building code requirements for
reinforced concrete, ACI 318-83.
Bazant, Z.P. and Oh. B.H.
(1976) Analysis of
(1984) Experimen-
materials like concrete. Report TVBM-1007, Div. of Building Materials, Lund Inst. of Technology, Sweden.
Petersson, P.E.